General scripting issues vs. Linux

[Backwoods BC]

On Fri, Mar 10, 2023 at 8:16 AM Brian Inglis via Cygwin
<cygwin@cygwin.com> wrote:
>
> On 2023-03-10 06:59, Ken Brown via Cygwin wrote:
> > On 3/10/2023 8:47 AM, Markus Becker via Cygwin wrote:
> >> I am quite an newby in Cygwin scripting and encountered several execution
> >> issues with bash scripts. For example, when i try to execute the following
> >> simple scriptfile "skript1.sh":
> >> # This is a testscript
> >> Statement="This is the testscript number 3"
> >> FILE="home/mbecker/Secure_Copy_Beispiel.txt"
> >> ls -l $FILE
> >> echo $Statement
> >> echo The file is $FILE
> >> i got these results:
> >> $ ./skript1.sh
> >> ls: cannot access 'home/mbecker/Secure_Copy_Beispiel.txt'$'\r\r': No such
>                                                               ^^^^^
> >> file or directory
> >> This is the testscript number 3
> >> The file is home/mbecker/Secure_Copy_Beispiel.txt
> >> or another results from a different script:
> >> $ ./skript7.sh
> >> ./skript7.sh: line 3: $'clear\r': command not found
>                                  ^^^
> >> Dr▒cken sie beliebige Tasten und dann return
> >> ': not a valid identifierd: `TASTE
> >> These are just two of several issues coming up with bash scripting in
> >> Cygwin. Maybe this is merely a corse problem with my platform
> >> understanding. But why is Cygwin calling errors when performing standard
> >> Linux bash commands? Is it due to a different syntax? Or is it even simpler?
>
> > It looks like your scripts have CRLF line endings.
>
> Utilities such as coreutils, gawk, grep, sed, etc. Cygwin packages had Cygwin
> tweaks removed in 2017 to be compatible with Linux and other platforms in
> handling '\r' before newlines, except for the single exception of Cygwin text
> mounts, where '\r' may be stripped if a program opens a file from that mount in
> text "t" mode, and may be added on writes to a file in text "t" mode on a text
> mount:
>
>         https://www.cygwin.com/cygwin-ug-net/using-textbinary.html
>
> discussions:
>
>         https://cygwin.com/legacy-ml/cygwin/2017-02/msg00152.html
>         https://cygwin.com/legacy-ml/cygwin/2017-02/msg00188.html
>         https://cygwin.com/legacy-ml/cygwin/2017-02/msg00189.html
>
> Install package dos2unix which conveniently strips the offending junk from your
> scripts and files e.g.
>
>         d2u -k skript*.sh
>
> Install and use Cygwin editors and utilities, or check editor and utility
> settings to ensure they are not set to behave like Windows e.g. gvim set
> fileformat=unix termencoding=utf-8 fileencoding=utf-8 in your ~/.gvimrc ~/.vimrc
> ~/.virc ~/.exrc; emacs (set-buffer-file-coding-system 'mule-utf-8-unix) in
> ~/.emacs, type C-x C-q C-m f mule-utf-8-unix, or do the equivalent in more
> sophisticated initializations (auto)detecting file type, encoding, and format;
> for git config --global core.autocrlf = input, see:
>
> https://stackoverflow.com/questions/3206843/how-line-ending-conversions-work-with-git-core-autocrlf-between-different-operat
>
> --
> Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada

As a matter of expediency, I always pass files through '| tr -d '\r'
|' when reading them. This is harmless on Linux and you don't have to
worry whether or not your target system has 'd2u' installed since I
believe that 'tr' is part of all base installations. It also means you
don't have to worry about whether or not a particular program handles
'\r' automagically.

I find that it is also a good idea to save all script files as UTF-8
so that they can handle Windows filenames with emojis and such in them
when explicitly named in the script.